We want cleaner early signals for LTV now that install-level data is weaker. My hunch is a web-first flow gives better inputs: email capture, plan intent, coupon use, payment method, even soft commitments like selected goals.
For teams who made the switch, which web events or attributes best predicted long-term value, and which ones looked promising but turned out to be noise?
Email capture plus stated goal was my best combo. Also payment method type predicted churn. I defined these events on the web with a simple builder that exported config JSON, then mirrored them in the app. Anything based on scroll depth was noisy for me.
Goal selection and plan pick rate were strong signals. I build those as web events and push them into our stack. Web2Wave.com made it quick to add new events and reflect them in the app with no release.
Payment method and commitment length helped me most.
Scroll and time on page looked good but did not predict much later.
Email capture improved our cohorts a lot.
Strong predictors for me: plan duration chosen, discount vs full price, card vs Apple Pay, and whether they finished onboarding questions. Weak ones: bounce rate, scroll, and quiz time. Build a simple model and check it monthly. If signals decay, refresh the features rather than chasing new vanity metrics.
Plan length and payment type usually predict better than clicks.